The threshold feeding response of microzooplankton within Pacific high-nitrate low-chlorophyll ecosystem models under steady and variable iron input

Abstract

The equatorial Pacific is an HNLC (High-Nitrate Low-Chlorophyll) region. Modeling and in-situ process studies have confirmed the importance of microzooplankton grazing in this ecosystem. Unfortunately, both the parameters and functions representing microzooplankton grazing within current ecosystem models are poorly constrained. We used a simple 4-component food web model to test the assumption that a lower grazing threshold, which is common in many models, is necessary to achieve the HNLC condition. Without the grazing threshold, the model did not reproduce the HNLC condition. However, by raising the half-saturation constant within the microzooplankton functional response with no threshold, it was possible to reproduce the critical dynamics of the HNLC condition under both steady and moderate seasonal variability in nutrient input. It was also possible to reproduce the HNLC system using a sigmoidal functional response for the microzooplankton, with results somewhere between the other two forms of the model, although this version had the highest sensitivity to changes in its parameters. The three models predicted similar phytoplankton biomass and primary productivity under steady nutrient input, but diverge in these metrics as the amplitude of nutrient input variability increases. These three functional responses also imply certain important differences in the microzooplankton community. Whereas the threshold model had the least sensitivity to parameter choice, the high half-saturation constant, no-threshold model may actually be a better approximation when modeling a community of grazers. Ecosystem models that predict carbon production and export in HNLC regions can be very sensitive to assumptions concerning microzooplankton grazing; future studies need to concentrate on the functional responses of microzooplankton before these models can be used for predicting fluxes in times or regions where forcing is beyond that used to constrain the original model.

Introduction

Large portions of the equatorial and oceanic pacific have been classified as being high-nitrate low-chlorophyll (HNLC) regimes (Martin et al., 1991; Chisholm and Morel, 1991; Frost and Kishi, 1999). Both field process studies, such as EqPac (Murray et al., 1994; Landry et al., 1995a; Landry et al., 1995b; Verity et al., 1996), and modeling studies (Frost and Franzen, 1992; Loukos et al., 1997; Pitchford and Brindley, 1999) have confirmed the importance of the microbial loop in the structure and maintenance of the HNLC condition. The current paradigm is that the growth rates of large phytoplankton are limited by the rate of nutrient input (iron), whereas growth and standing stock of small phytoplankton biomass, which constitutes the majority of total phytoplankton, is limited through tight grazing control by the microzooplankton and recycling efficiency (Banse, 1992; Frost and Franzen, 1992; Landry et al., 1997). Because of this tight coupling between the microzooplankton and phytoplankton, a critical component of any ecosystem model describing an HNLC system is the form of the functional response used to represent the grazing of the microzooplankton community on the phytoplankton community.

Many ecosystem models typically use a Michaelis-Menten or similar Type II (Holling, 1959) function with some kind of grazing threshold (Fig. 1) to describe the functional response of zooplankton grazing. The grazing threshold, or “lower feeding threshold,” is defined as the concentration of prey, below which, the predator stops feeding. An alternative is to use a Type III (sigmoidal, Holling, 1959) functional response, which, due to the inflection in ingestion rate as the prey concentration decreases, acts the same as a threshold model (Steele, 1974a; Steele and Mullin, 1977). Steele 1974a, Steele 1974b first pointed out the importance of the grazing function in controlling the dynamics of a planktonic food web, and suggested that a lower feeding threshold is actually critical to stabilize these models. Essentially, without the lower threshold it is assumed that predators can completely eliminate their prey, which is not likely in nature. The other possibilities that result when a threshold is not included are chaotic fluctuations or stable limit cycles (Steele and Henderson, 1992). The existence of a lower grazing threshold for individual zooplankton predators feeding on a single prey type has both theoretical (Steele, 1974b; Strom et al., 2000) and experimental (some of which is equivocal) support (Frost, 1974; Rivier et al., 1985; Verity, 1991; Choi, 1994; Lessard and Murrell, 1998). However, the existence of this threshold has never been confirmed for the specific microzooplankton assemblage of the equatorial Pacific HNLC.

An “apparent” threshold also can be observed, even if an individual zooplankton species does not have a grazing threshold, if multiple prey types are present (Fasham et al., 1990). This is because when two (or more) acceptable prey types are present, a predator should eat proportionally more of the more abundant prey type, or the predator may switch between prey items. Thus, the shape of the functional response of a predator on one individual prey item could show either a threshold or sigmoidal shape at low prey concentrations (Fig. 1). Franks et al. (1986) showed that inclusion of a sigmoidal Type III-like function (the function was Type III only at lower concentrations of prey) provided model stability in a simple nitrogen–phytoplankton–zooplankton (NPZ) model. Fasham (1995) also found that a Type III model resulted in greater model stability in a somewhat more complicated ecosystem model. Several ecosystem models have taken this approach to modeling an HNLC food web; i.e. they include no lower feeding threshold, but rather a “perceived” threshold on a single prey item due to the availability of multiple prey (Fasham et al., 1990; Loukos et al., 1997; Pitchford and Brindley, 1999). Thus, ecosystem models of HNLC regimes so far have always included either a threshold (or Type III response) or prey-switching ability of the microzooplankton in order to achieve the HNLC condition; without these assumptions in the model structure, models may not produce an HNLC regime (Strom et al., 2000). However, the particular formulation used to model prey switching is very critical, as most commonly used formulations contain both mathematical and conceptual inconsistencies (Gentleman et al., 2003).

One of the stated goals of the Joint Global Ocean Flux Study (JGOFS) is to develop a simple useful model that matches local data well, and can eventually be extrapolated to other areas (Evans, 1999). Before models that are fit to local data can do this, the potential impacts of the assumptions described above must be more thoroughly addressed for the microzooplankton inhabiting the HNLC equatorial Pacific. Addressing these assumptions concerning the grazing function is doubly important for both increasing the predictive ability of such models when coupled to larger-scale physical models, and to increase our basic mechanistic understanding of how the food web functions. Here, we tested the hypothesis that inclusion of a lower feeding threshold is necessary to achieve an HNLC system. We also compared the sensitivity of daily primary production and phytoplankton standing stock within different forms of a microzooplankton grazing response (threshold versus no threshold versus Type III). The overall goal of this work was to analyze the appropriateness of the microzooplankton grazing functions currently used in most ecosystem models, and address their impact (faulty or otherwise) on predictions of primary production and carbon export made from such models. In a later contribution, we shall address the second assumption implicit in many ecosystem models; the formulation used for feeding on multiple prey types (Gentleman et al., 2003).